University of Economics, Prague
The early-stage design of a new microprocessor involves the evaluation of a wide range of benchmarks across a large number of architectural configurations. Several methods are used to cut down on the required simulation time. Typically, however, existing approaches fail to capture true program behavior accurately and require a non-negligible number of training simulations to be run. The authors address these problems by developing a machine learning model that predicts the mean of any given metric, e.g. cycles or energy, across a range of programs, for any microarchitectural configuration.